Queue machines: An organization for parallel computation
We explore parallel system organizations based on a representation of programs which allows execution using a queue as a working store. The main advantage of such a representation, called Q-notation, is that multiple processors can be used in a very regular manner, so that a simple and natural mapping of parallel computations onto parallel processors is achieved. The proposed machines are characterized by very efficient and fast instruction issue, modularity with useful fault-tolerance properties, and simplified interconnection requirements. We define a Q-notation for program representation and discuss in general its capabilities and limitations in executing ordinary (sequential) and concurrent programs.
Unable to display preview. Download preview PDF.
- (1).Arvind, "Decomposing a Program for Multiple Processor Systems," International Conference on Parallel Processing, 1980.Google Scholar
- (4).Dennis, J.B., "The Avrieties of Data Flow Computers," Proc. International Conference on Distributed Systems, Huntsville, Alabama, 1979.Google Scholar
- (5).Feller, M., "A Parallel Queue Organization for High-Speed Computing," (thesis) UCLA, Los Angeles, California, 1980.Google Scholar
- (6).Flynn, M. J. and J. L. Hennessy, "Parallelism and Representation Problems in Distributed Systems," Proc. International Conference on Distributed Systems, Huntsville, Alabama, 1979.Google Scholar
- (7).Keller, R.M., G. Lindstrom, and S. Patil, "A Loosely-Coupled Applicative Multiprocessor System," Proc. AFIPS NCC, Vol. 48 (June 1979), 613–622.Google Scholar
- (9).Lang, T., "Interconnections Between Processors and Memory Modules Using the Shuffle-Exchange Network," IEEE Trans. Comput. C-25:5 (May 1976), 496–503.Google Scholar
- (10).Pawlak, Z., "New Class of Mathematical Languages and Organization of Addressless Computers," (in) Colloquium on the Foundation of Mathematics, Budapest: Akademiai Kiado, 1965, 1965, 227–238.Google Scholar
- (11).Stone, H.S., "A Pipeline Push-Down Stack Computer," (in) L.C. Hobbs et al., eds., Parallel Processor Systems, Technologies, and Applications. New York: Spartan 1970, 235–249.Google Scholar